I know I give #LLM a lot of hate. It's mostly deserved. But one area where #ChatGPT excels is helping to provide the names of things. So I asked chatsy (what @Jillianmarisa calls it) to name the hole on back of lawn tractors for the hitch pin for towing accessories. BC you can't research solutions without knowing what the darned thing is called. It's the pin hole which is part of the hitch plate. Checks out. Yay...
GPT-5 too unfriendly: OpenAI reverts to 4o as standard model
After one week of GPT-5, OpenAI responds to criticism: paying users get 4o back as standard. GPT-5 routing now also works in manual mode.
Das #GPT5 keine Karten zeichnen kann, ist übrigens Absicht.
Jetzt wisst ihr Bescheid. 
Gemini ganha 'dupla personalidade' com modo incógnito e uma nova memória
https://tugatech.com.pt/t70596-gemini-ganha-dupla-personalidade-com-modo-incognito-e-uma-nova-memoria
#OpenAI is rolling out support for #Gmail, #GoogleCalendar, and #GoogleContacts #Connectors in #ChatGPT (Pro). This allows ChatGPT to reference information from these services in conversations, making it faster and easier to bring information into chats. https://www.bleepingcomputer.com/news/artificial-intelligence/openai-rolls-out-gmail-calendar-and-contacts-integration-in-chatgpt/?eicker.news #tech #media #news
Cómo crear personajes que hablan individualmente con una única herramienta (HeyGen)
Una guía paso a paso para aprender a crear un personaje hablando con sincronización labial y movimientos, con la herramienta HeyGen.
Updated the #ChatGPT version with elasticity and variable number of veggies too: https://chatgpt.com/canvas/shared/689ca0c6bb74819180f61ca91ee8af2f
Santander une-se à OpenAI para criar um banco “nativo de IA” https://tugatech.com.pt/t70582-santander-une-se-a-openai-para-criar-um-banco-nativo-de-ia
Companies Are Pouring Billions Into A.I. It Has Yet to Pay Off. - Corporate spending on artificial intelligence is surging as executives bank on major effi... - https://www.nytimes.com/2025/08/13/business/ai-business-payoff-lags.html #computersandtheinternet #artificialintelligence #laborandjobs #innovation #start-ups #research #chatgpt
Back in the early days of the Internet when ordering pizza online was new, a pizza site's brilliant web dev coded a cursor with vegetables dangling on a string. When you moved the cursor, the veggies followed the laws of gravity and momentum. I've never found that or an equivalent site since so I asked #ChatGPT and #Gemini to create a simulation. Gemini failed miserably but after just a few tweaks, ChatGPT nailed it. Here's a demo. It's stunning how easy it was to do. 
https://chatgpt.com/canvas/shared/689ca0c6bb74819180f61ca91ee8af2f
OpenAI’s terms: “Our Services are not intended for use in the diagnosis or treatment of any health condition.”
Sam Altman in the GPT-5 release: “You’ve all seen examples of people getting day-to-day care advice, or sometimes even a lifesaving diagnosis [from ChatGPT].”
Experts warn that this fast-and-loose approach could land OpenAI in hot water w the FDA.
More:
https://www.statnews.com/2025/08/13/openai-cant-have-it-both-ways-on-gpt-5-and-health-ai-prognosis/
"Chatbots can privilege staying in character over following the safety guardrails that companies have put in place." —Kashmir Hill and Dylan Freedman for The New York Times
https://www.nytimes.com/2025/08/08/technology/ai-chatbots-delusions-chatgpt.html/?src=longreads
#AI companies claim #LLMs pass tests of medical knowledge with 80% accuracy. new study finds that in non-trivial cases it's actually more like 42%.
the main issue, of course, is that 80% or 40%, the bot and the megacorps behind them are always 100% confident.
you are being lied to. a dangerous lie that could be fatal.
Fidelity of Medical Reasoning in Large Language Models | Medical Education and Training | JAMA Network Open | JAMA Network
https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2837372
Every once in a while I ask a #virtual #dumbass (LLMs, such as #Copilot or #ChatGPT) for an information that I need.
The results are impressive! The amount of useful answers I got was exactly: 0. In words: zero. That's almost 0%.
The horrible part about this is that people use those bullshit generators to get actual work "done", not realizing, that they just got a bunch of BS.
"AI will soon take over the World!!!11"
lol, no.
You know what's great?
Welcome to a world where people are so afraid of thinking they use #ChatGPT to generate better prompts for their image generators.
GPT-5 im Vergleich: Softwareentwicklung sehr gut, Kreativität nur ausreichend
OpenAIs frisch veröffentlichtes großes Sprachmodell GPT-5 ist effizient und hat einige Stärken, bleibt aber an manchen Stellen hinter der Konkurrenz zurück.
Wurden diese 4400 offenbar überflüssigen Lehrystellen in BaWü mittlerweile schon preiswert durch PaukGPT ersetzt?
On Aug 7, 2025, OpenAI launched #GPT5, featuring a router that decides when to “think” longer, new model sizes, and pricing designed for production use.
Highlights: 400K token context + 128K max output tokens.
Learn more: https://bit.ly/4otArXv
Gus Posey 
